Explaratory Data Analysis
The goal of the exploratory analysis is to identify the key predictors, generate insights and dicover trends and patters. Throughout this process we are mainly interested in detecting:
- Correlation and assoicitations
- Trends in time e.g viral load
- Outliers
- Any other key information that might be useful in ML modeling
Assumptions
- AMPATH civil status is collected longitudinal; as such missing data on the crossectional dataset has to be replaced with longitudinal civil status recorded not > 3 months after the first encounter
- We have assumed if is.na(transfer) then No else Yes
- Unrecorded data (occurring at the 1st encounter) on Height, Weight, BMI, NHIF have to be replaced with non-null next encounter recorded < 3 months from the first encounter
- Viral Load index - It is not possible to know when the next VL is drawn longitudinally, as shown in the table below. As such, we are treating each VL result as individual tests ordered independently.
| ptidno | visit_num | apptdate | VLundetectable | viralload | viralloadc | Height |
|---|---|---|---|---|---|---|
| 52764 | 1 | 2017-01-01 | NA | NA | NA | |
| 52764 | 2 | 2017-04-25 | NA | NA | 162 | |
| 52764 | 3 | 2017-05-24 | NA | NA | 161 | |
| 52764 | 4 | 2017-06-28 | NA | NA | 162 | |
| 52764 | 5 | 2017-07-21 | NA | 40 | < | 162 |
| 52764 | 6 | 2017-08-22 | NA | NA | 162 | |
| 52764 | 7 | 2017-09-12 | NA | NA | 162 | |
| 52764 | 8 | 2017-10-16 | NA | NA | 162 | |
| 52764 | 9 | 2017-11-20 | NA | NA | 162 | |
| 52764 | 10 | 2017-12-18 | NA | NA | 162 | |
| 52764 | 11 | 2018-01-01 | NA | 40 | < | 162 |
| 52764 | 12 | 2018-02-19 | NA | NA | 162 | |
| 52764 | 13 | 2018-03-06 | NA | NA | 162 | |
| 52764 | 14 | 2018-05-07 | NA | 40 | < | 162 |
| 52764 | 15 | 2018-06-05 | NA | NA | 162 | |
| 52764 | 16 | 2018-09-10 | NA | NA | 162 | |
| 52764 | 17 | 2018-12-04 | NA | 40 | < | 162 |
| 52764 | 18 | 2019-01-01 | NA | NA | 162 | |
| 52764 | 19 | 2019-04-26 | NA | NA | 173 | |
| 52764 | 20 | 2019-07-17 | NA | NA | 173 | |
| 52764 | 21 | 2019-10-11 | NA | 40 | < | NA |
| 52764 | 22 | 2019-10-14 | NA | NA | NA | |
| 52764 | 23 | 2019-11-13 | NA | NA | NA |
Viral Load Counts
We will start by looking at the overall distribution of VL
Distribution
Is a cutoff of 1000 valid with relation to VL detectability limit? To understll start by decomposing the viral load counts into its three components that is inline with the goal of our analyis- Viral load 1, viral load 2 and viral load 3.
Overall Viral Load Counts Detection Limit
MBARARA Viral Load Detection Limit
MASAKA Viral Load Detection Limit
RAKAI Viral Load Detection Limit
IDI Viral Load Detection Limit
AMPATH Viral Load Detection Limit
FACES Viral Load Detection Limit
KISESA Viral Load Detection Limit
MOROGORO Viral Load Detection Limit
TUMBI Viral Load Detection Limit
We also need to look if this is consisten with VL index
Viral Load 1 Detection Limit
Viral Load 2 Detection Limit
Viral Load 3 Detection Limit
First VL Seasonality Trends
Plot the explanatory variable distribution for both the variables to understand the variability uniquely explained (The non-intersecting part of the blue and the red is the variation explained by the variable)
Missing Data
Socio-demographics Missing Data Summary
| variable | n_miss | pct_miss |
|---|---|---|
| nhif | 83396 | 96.147018 |
| vl_count3 | 68926 | 79.464595 |
| vl_count2 | 60657 | 69.931287 |
| vl_count1 | 48470 | 55.880929 |
| entrypoint | 46852 | 54.015541 |
| civilstatusENROL | 22158 | 25.545897 |
| bmi | 19876 | 22.914985 |
| Height | 19581 | 22.574881 |
| ageatarvstart | 11592 | 13.364385 |
| Weight | 6309 | 7.273629 |
| male | 0 | 0.000000 |
| Death | 0 | 0.000000 |
| transfer | 0 | 0.000000 |
| program | 0 | 0.000000 |
Bivariate Analysis
To understand the relationship of each variable with the outcome variable:
Socio-demographics
Overall Socio-demographics Correlation
Socio-demographics | Viral Load Failure 1
| Table 2: Socio-demographics | ||||||
| First Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 38,268 |
 |
No No. 32,723 |
Yes No. 5,545 |
 | P-value | |
| age at first exposure to ARV | 0.49 | |||||
|   Mean (SD) | 36.2 (±10.8) |  | 36.2 (±10.8) | 36.3 (±10.8) |  | |
| Â Â Median (IQR) | 34.5 (28.1 - 42.7) | Â | 34.5 (28.1 - 42.6) | 34.6 (28.1 - 42.9) | Â | |
| Â Â Missing | 685 (1.8%) | Â | 549 (1.7%) | 136 (2.5%) | Â | |
| male gender | 0.004 | |||||
| Â Â No | 24,425 (63.8%) | Â | 20,982 (85.9%) | 3,443 (14.1%) | Â | |
| Â Â Yes | 13,843 (36.2%) | Â | 11,741 (84.8%) | 2,102 (15.2%) | Â | |
| death | < 0.0001 | |||||
| Â Â No | 37,473 (97.9%) | Â | 32,269 (86.1%) | 5,204 (13.9%) | Â | |
| Â Â Yes | 795 (2.1%) | Â | 454 (57.1%) | 341 (42.9%) | Â | |
| civil status at enrollment | 0.002 | |||||
| Â Â Divorced | 716 (1.9%) | Â | 600 (83.8%) | 116 (16.2%) | Â | |
| Â Â Legally Married | 18,280 (47.8%) | Â | 15,630 (85.5%) | 2,650 (14.5%) | Â | |
| Â Â Living w/Partner | 501 (1.3%) | Â | 433 (86.4%) | 68 (13.6%) | Â | |
| Â Â Never Married and Not Living w/Partner | 4,098 (10.7%) | Â | 3,414 (83.3%) | 684 (16.7%) | Â | |
| Â Â Separated | 4,348 (11.4%) | Â | 3,659 (84.2%) | 689 (15.8%) | Â | |
| Â Â Widowed | 2,918 (7.6%) | Â | 2,445 (83.8%) | 473 (16.2%) | Â | |
| Â Â Missing | 7,407 (19.4%) | Â | 6,542 (88.3%) | 865 (11.7%) | Â | |
| patient transferred out | < 0.0001 | |||||
| Â Â No | 36,344 (95.0%) | Â | 31,187 (85.8%) | 5,157 (14.2%) | Â | |
| Â Â Yes | 1,924 (5.0%) | Â | 1,536 (79.8%) | 388 (20.2%) | Â | |
| Height (cm) at enrollment | 0.060 | |||||
|   Mean (SD) | 164.5 (±8.9) |  | 164.4 (±8.9) | 164.7 (±8.9) |  | |
| Â Â Median (IQR) | 164.0 (158.0 - 170.0) | Â | 164.0 (158.0 - 170.0) | 164.0 (158.5 - 170.5) | Â | |
| Â Â Missing | 7,610 (19.9%) | Â | 6,497 (19.9%) | 1,113 (20.1%) | Â | |
| Weight (kg) at enrollment | < 0.0001 | |||||
|   Mean (SD) | 58.6 (±11.6) |  | 59.0 (±11.6) | 56.8 (±11.0) |  | |
| Â Â Median (IQR) | 57.0 (51.0 - 65.0) | Â | 57.7 (51.0 - 65.0) | 55.5 (50.0 - 63.0) | Â | |
| Â Â Missing | 1,550 (4.1%) | Â | 1,300 (4.0%) | 250 (4.5%) | Â | |
| BMI at enrollment | < 0.0001 | |||||
|   Mean (SD) | 21.8 (±4.2) |  | 21.9 (±4.2) | 20.9 (±3.9) |  | |
| Â Â Median (IQR) | 21.1 (18.9 - 23.8) | Â | 21.2 (19.1 - 23.9) | 20.4 (18.3 - 22.9) | Â | |
| Â Â Missing | 7,698 (20.1%) | Â | 6,566 (20.1%) | 1,132 (20.4%) | Â | |
| Second VL count | < 0.0001 | |||||
|   Mean (SD) | 14136.7 (±161008.3) |  | 4297.9 (±88387.6) | 69708.8 (±353005.0) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 283.0) | Â | 40.0 (40.0 - 128.0) | 481.0 (40.0 - 19310.5) | Â | |
| Â Â Missing | 12,187 (31.8%) | Â | 10,565 (32.3%) | 1,622 (29.3%) | Â | |
| Third VL count | < 0.0001 | |||||
|   Mean (SD) | 13619.4 (±843866.8) |  | 2780.0 (±36129.8) | 69173.1 (±2086307.2) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 197.0) | Â | 40.0 (40.0 - 112.0) | 135.5 (40.0 - 2376.8) | Â | |
| Â Â Missing | 20,456 (53.5%) | Â | 17,819 (54.5%) | 2,637 (47.6%) | Â | |
| ‡ | ||||||
Socio-demographics | Viral Load Failure 2
| Table 2: Socio-demographics | ||||||
| Second Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 26,081 |
 |
No No. 22,722 |
Yes No. 3,359 |
 | P-value | |
| age at first exposure to ARV | 0.068 | |||||
|   Mean (SD) | 36.7 (±10.8) |  | 36.7 (±10.8) | 37.0 (±10.8) |  | |
| Â Â Median (IQR) | 35.1 (28.5 - 43.3) | Â | 35.1 (28.5 - 43.2) | 35.5 (28.7 - 43.8) | Â | |
| Â Â Missing | 332 (1.3%) | Â | 294 (1.3%) | 38 (1.1%) | Â | |
| male gender | < 0.0001 | |||||
| Â Â No | 16,797 (64.4%) | Â | 14,745 (87.8%) | 2,052 (12.2%) | Â | |
| Â Â Yes | 9,284 (35.6%) | Â | 7,977 (85.9%) | 1,307 (14.1%) | Â | |
| death | < 0.0001 | |||||
| Â Â No | 25,715 (98.6%) | Â | 22,505 (87.5%) | 3,210 (12.5%) | Â | |
| Â Â Yes | 366 (1.4%) | Â | 217 (59.3%) | 149 (40.7%) | Â | |
| civil status at enrollment | 0.002 | |||||
| Â Â Divorced | 423 (1.6%) | Â | 352 (83.2%) | 71 (16.8%) | Â | |
| Â Â Legally Married | 13,138 (50.4%) | Â | 11,411 (86.9%) | 1,727 (13.1%) | Â | |
| Â Â Living w/Partner | 313 (1.2%) | Â | 277 (88.5%) | 36 (11.5%) | Â | |
| Â Â Never Married and Not Living w/Partner | 2,764 (10.6%) | Â | 2,338 (84.6%) | 426 (15.4%) | Â | |
| Â Â Separated | 2,978 (11.4%) | Â | 2,607 (87.5%) | 371 (12.5%) | Â | |
| Â Â Widowed | 2,218 (8.5%) | Â | 1,904 (85.8%) | 314 (14.2%) | Â | |
| Â Â Missing | 4,247 (16.3%) | Â | 3,833 (90.3%) | 414 (9.7%) | Â | |
| patient transferred out | < 0.0001 | |||||
| Â Â No | 25,078 (96.2%) | Â | 21,900 (87.3%) | 3,178 (12.7%) | Â | |
| Â Â Yes | 1,003 (3.8%) | Â | 822 (82.0%) | 181 (18.0%) | Â | |
| Height (cm) at enrollment | 0.009 | |||||
|   Mean (SD) | 164.7 (±8.9) |  | 164.7 (±8.9) | 165.1 (±8.9) |  | |
| Â Â Median (IQR) | 164.0 (159.0 - 170.3) | Â | 164.0 (158.5 - 170.0) | 165.0 (159.0 - 171.0) | Â | |
| Â Â Missing | 5,383 (20.6%) | Â | 4,686 (20.6%) | 697 (20.8%) | Â | |
| Weight (kg) at enrollment | < 0.0001 | |||||
|   Mean (SD) | 58.9 (±11.6) |  | 59.2 (±11.7) | 57.2 (±10.9) |  | |
| Â Â Median (IQR) | 57.8 (51.0 - 65.0) | Â | 58.0 (51.0 - 65.0) | 56.0 (50.0 - 63.0) | Â | |
| Â Â Missing | 1,006 (3.9%) | Â | 859 (3.8%) | 147 (4.4%) | Â | |
| BMI at enrollment | < 0.0001 | |||||
|   Mean (SD) | 21.8 (±4.2) |  | 21.9 (±4.2) | 21.0 (±3.9) |  | |
| Â Â Median (IQR) | 21.1 (19.0 - 23.9) | Â | 21.2 (19.0 - 24.0) | 20.4 (18.4 - 22.9) | Â | |
| Â Â Missing | 5,439 (20.9%) | Â | 4,733 (20.8%) | 706 (21.0%) | Â | |
| First VL count | < 0.0001 | |||||
|   Mean (SD) | 17759.0 (±168420.6) |  | 8511.1 (±128813.3) | 80316.6 (±321771.5) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 400.0) | Â | 40.0 (40.0 - 213.8) | 1222.0 (40.0 - 33766.5) | Â | |
| Third VL count | < 0.0001 | |||||
|   Mean (SD) | 13619.4 (±843866.8) |  | 2244.6 (±33820.5) | 81018.8 (±2218370.9) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 197.0) | Â | 40.0 (40.0 - 100.0) | 400.0 (40.0 - 9315.2) | Â | |
| Â Â Missing | 8,269 (31.7%) | Â | 7,482 (32.9%) | 787 (23.4%) | Â | |
| ‡ | ||||||
Socio-demographics | Viral Load Failure 3
| Table 3: Socio-demographics | ||||||
| Third Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 17,812 |
 |
No No. 15,776 |
Yes No. 2,036 |
 | P-value | |
| age at first exposure to ARV | 0.69 | |||||
|   Mean (SD) | 37.2 (±10.8) |  | 37.2 (±10.8) | 37.2 (±10.9) |  | |
| Â Â Median (IQR) | 35.8 (29.0 - 44.0) | Â | 35.8 (29.0 - 44.0) | 35.8 (28.9 - 43.8) | Â | |
| Â Â Missing | 166 (0.9%) | Â | 147 (0.9%) | 19 (0.9%) | Â | |
| male gender | < 0.0001 | |||||
| Â Â No | 11,510 (64.6%) | Â | 10,308 (89.6%) | 1,202 (10.4%) | Â | |
| Â Â Yes | 6,302 (35.4%) | Â | 5,468 (86.8%) | 834 (13.2%) | Â | |
| death | < 0.0001 | |||||
| Â Â No | 17,637 (99.0%) | Â | 15,669 (88.8%) | 1,968 (11.2%) | Â | |
| Â Â Yes | 175 (1.0%) | Â | 107 (61.1%) | 68 (38.9%) | Â | |
| civil status at enrollment | 0.12 | |||||
| Â Â Divorced | 270 (1.5%) | Â | 237 (87.8%) | 33 (12.2%) | Â | |
| Â Â Legally Married | 9,568 (53.7%) | Â | 8,456 (88.4%) | 1,112 (11.6%) | Â | |
| Â Â Living w/Partner | 197 (1.1%) | Â | 176 (89.3%) | 21 (10.7%) | Â | |
| Â Â Never Married and Not Living w/Partner | 1,828 (10.3%) | Â | 1,578 (86.3%) | 250 (13.7%) | Â | |
| Â Â Separated | 2,025 (11.4%) | Â | 1,805 (89.1%) | 220 (10.9%) | Â | |
| Â Â Widowed | 1,694 (9.5%) | Â | 1,505 (88.8%) | 189 (11.2%) | Â | |
| Â Â Missing | 2,230 (12.5%) | Â | 2,019 (90.5%) | 211 (9.5%) | Â | |
| patient transferred out | < 0.0001 | |||||
| Â Â No | 17,275 (97.0%) | Â | 15,334 (88.8%) | 1,941 (11.2%) | Â | |
| Â Â Yes | 537 (3.0%) | Â | 442 (82.3%) | 95 (17.7%) | Â | |
| Height (cm) at enrollment | 0.050 | |||||
|   Mean (SD) | 165.1 (±8.9) |  | 165.0 (±8.9) | 165.4 (±8.8) |  | |
| Â Â Median (IQR) | 165.0 (159.0 - 171.0) | Â | 165.0 (159.0 - 171.0) | 165.0 (159.5 - 171.0) | Â | |
| Â Â Missing | 3,863 (21.7%) | Â | 3,434 (21.8%) | 429 (21.1%) | Â | |
| Weight (kg) at enrollment | < 0.0001 | |||||
|   Mean (SD) | 59.3 (±11.8) |  | 59.5 (±11.9) | 57.2 (±10.7) |  | |
| Â Â Median (IQR) | 58.0 (51.0 - 65.0) | Â | 58.0 (51.0 - 66.0) | 56.0 (50.0 - 63.0) | Â | |
| Â Â Missing | 686 (3.9%) | Â | 589 (3.7%) | 97 (4.8%) | Â | |
| BMI at enrollment | < 0.0001 | |||||
|   Mean (SD) | 21.8 (±4.2) |  | 21.9 (±4.3) | 20.9 (±3.8) |  | |
| Â Â Median (IQR) | 21.1 (18.9 - 23.9) | Â | 21.2 (19.0 - 24.0) | 20.3 (18.4 - 22.9) | Â | |
| Â Â Missing | 3,901 (21.9%) | Â | 3,465 (22.0%) | 436 (21.4%) | Â | |
| First VL count | < 0.0001 | |||||
|   Mean (SD) | 17853.8 (±161750.7) |  | 12629.9 (±145863.6) | 58330.8 (±249418.1) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 400.0) | Â | 40.0 (40.0 - 316.0) | 522.0 (40.0 - 14620.0) | Â | |
| Second VL count | < 0.0001 | |||||
|   Mean (SD) | 14532.3 (±161359.8) |  | 7872.0 (±115264.2) | 66139.5 (±349122.0) |  | |
| Â Â Median (IQR) | 40.0 (40.0 - 380.2) | Â | 40.0 (40.0 - 192.0) | 1367.5 (40.0 - 21434.8) | Â | |
| ‡ | ||||||
Socio-demographics | Program
| Table 4: Socio-demographics | |||||||||||
| Programs | |||||||||||
|---|---|---|---|---|---|---|---|---|---|---|---|
|
Total No. 86,738 |
 |
AMPATH No. 42,756 |
FACES No. 10,329 |
IDI No. 2,244 |
KISESA No. 1,306 |
MASAKA No. 4,267 |
MBARARA No. 2,963 |
MOROGORO No. 1,785 |
RAKAI No. 19,688 |
TUMBI No. 1,400 |
|
| age at first exposure to ARV | |||||||||||
|   Mean (SD) | 35.1 (±10.9) |  | 37.0 (±11.2) | 32.4 (±9.8) | 35.6 (±10.8) | 35.2 (±11.0) | 31.7 (±9.3) | 31.7 (±9.7) | 40.0 (±11.4) | 32.5 (±9.9) | 39.6 (±11.7) |
| Â Â Median (IQR) | 33.0 (26.9 - 41.3) | Â | 35.4 (28.6 - 43.8) | 30.4 (25.3 - 37.2) | 33.6 (27.7 - 42.0) | 33.1 (27.4 - 40.6) | 30.0 (24.7 - 37.1) | 29.4 (24.5 - 36.2) | 39.3 (31.6 - 47.1) | 30.4 (25.0 - 37.9) | 38.7 (30.7 - 46.0) |
| Â Â Missing | 11,592 (13.4%) | Â | 5,479 (12.8%) | 1,666 (16.1%) | 312 (13.9%) | 78 (6.0%) | 147 (3.4%) | 123 (4.2%) | 43 (2.4%) | 3,679 (18.7%) | 65 (4.6%) |
| male gender | |||||||||||
| Â Â No | 54,332 (62.6%) | Â | 27,028 (49.7%) | 6,496 (12.0%) | 1,316 (2.4%) | 780 (1.4%) | 2,673 (4.9%) | 1,836 (3.4%) | 1,097 (2.0%) | 12,224 (22.5%) | 882 (1.6%) |
| Â Â Yes | 32,406 (37.4%) | Â | 15,728 (48.5%) | 3,833 (11.8%) | 928 (2.9%) | 526 (1.6%) | 1,594 (4.9%) | 1,127 (3.5%) | 688 (2.1%) | 7,464 (23.0%) | 518 (1.6%) |
| death | |||||||||||
| Â Â No | 82,158 (94.7%) | Â | 39,877 (48.5%) | 9,883 (12.0%) | 2,012 (2.4%) | 1,254 (1.5%) | 4,140 (5.0%) | 2,914 (3.5%) | 1,662 (2.0%) | 19,120 (23.3%) | 1,296 (1.6%) |
| Â Â Yes | 4,580 (5.3%) | Â | 2,879 (62.9%) | 446 (9.7%) | 232 (5.1%) | 52 (1.1%) | 127 (2.8%) | 49 (1.1%) | 123 (2.7%) | 568 (12.4%) | 104 (2.3%) |
| civil status at enrollment | |||||||||||
| Â Â Divorced | 1,877 (2.2%) | Â | 422 (22.5%) | 0 (0.0%) | 8 (0.4%) | 28 (1.5%) | 837 (44.6%) | 98 (5.2%) | 141 (7.5%) | 297 (15.8%) | 46 (2.5%) |
| Â Â Legally Married | 37,068 (42.7%) | Â | 20,314 (54.8%) | 6,091 (16.4%) | 1,078 (2.9%) | 808 (2.2%) | 1,377 (3.7%) | 1,650 (4.5%) | 280 (0.8%) | 5,242 (14.1%) | 228 (0.6%) |
| Â Â Living w/Partner | 1,017 (1.2%) | Â | 591 (58.1%) | 91 (8.9%) | 173 (17.0%) | 1 (0.1%) | 0 (0.0%) | 0 (0.0%) | 2 (0.2%) | 144 (14.2%) | 15 (1.5%) |
| Â Â Never Married and Not Living w/Partner | 8,941 (10.3%) | Â | 5,596 (62.6%) | 0 (0.0%) | 405 (4.5%) | 435 (4.9%) | 348 (3.9%) | 629 (7.0%) | 69 (0.8%) | 1,327 (14.8%) | 132 (1.5%) |
| Â Â Separated | 10,224 (11.8%) | Â | 5,224 (51.1%) | 2,414 (23.6%) | 430 (4.2%) | 0 (0.0%) | 0 (0.0%) | 466 (4.6%) | 0 (0.0%) | 1,690 (16.5%) | 0 (0.0%) |
| Â Â Widowed | 5,453 (6.3%) | Â | 3,777 (69.3%) | 803 (14.7%) | 140 (2.6%) | 32 (0.6%) | 45 (0.8%) | 118 (2.2%) | 38 (0.7%) | 463 (8.5%) | 37 (0.7%) |
| Â Â Missing | 22,158 (25.5%) | Â | 6,832 (30.8%) | 930 (4.2%) | 10 (0.0%) | 2 (0.0%) | 1,660 (7.5%) | 2 (0.0%) | 1,255 (5.7%) | 10,525 (47.5%) | 942 (4.3%) |
| patient transferred out | |||||||||||
| Â Â No | 80,287 (92.6%) | Â | 39,477 (49.2%) | 9,109 (11.3%) | 2,055 (2.6%) | 964 (1.2%) | 3,681 (4.6%) | 2,675 (3.3%) | 1,661 (2.1%) | 19,580 (24.4%) | 1,085 (1.4%) |
| Â Â Yes | 6,451 (7.4%) | Â | 3,279 (50.8%) | 1,220 (18.9%) | 189 (2.9%) | 342 (5.3%) | 586 (9.1%) | 288 (4.5%) | 124 (1.9%) | 108 (1.7%) | 315 (4.9%) |
| Weight (kg) at enrollment | |||||||||||
|   Mean (SD) | 57.8 (±11.3) |  | 58.4 (±11.5) | 59.7 (±10.9) | 59.9 (±13.7) | 56.0 (±10.2) | 57.1 (±10.7) | 60.0 (±12.0) | 58.3 (±13.7) | 54.9 (±9.4) | 58.7 (±13.6) |
| Â Â Median (IQR) | 56.5 (50.0 - 64.0) | Â | 57.0 (50.0 - 65.0) | 59.0 (52.0 - 65.0) | 58.0 (50.1 - 67.0) | 55.0 (50.0 - 61.0) | 55.8 (50.0 - 63.0) | 58.0 (51.0 - 66.0) | 56.0 (49.0 - 65.0) | 54.0 (49.0 - 60.0) | 57.0 (50.0 - 66.0) |
| Â Â Missing | 6,309 (7.3%) | Â | 3,426 (8.0%) | 131 (1.3%) | 90 (4.0%) | 2 (0.2%) | 91 (2.1%) | 89 (3.0%) | 29 (1.6%) | 2,388 (12.1%) | 63 (4.5%) |
| BMI at enrollment | |||||||||||
|   Mean (SD) | 21.5 (±4.1) |  | 21.2 (±4.1) | 21.4 (±3.8) | 22.8 (±5.2) | 20.8 (±3.7) | 21.9 (±4.0) | 23.1 (±4.6) | 23.1 (±5.3) | 21.4 (±3.6) | 22.1 (±5.3) |
| Â Â Median (IQR) | 20.9 (18.7 - 23.5) | Â | 20.6 (18.5 - 23.2) | 20.9 (18.9 - 23.3) | 21.8 (19.3 - 25.3) | 20.3 (18.4 - 22.7) | 21.3 (19.2 - 23.8) | 22.4 (20.0 - 25.5) | 22.2 (19.6 - 25.6) | 21.0 (19.1 - 23.2) | 21.3 (18.5 - 24.7) |
| Â Â Missing | 19,876 (22.9%) | Â | 9,227 (21.6%) | 158 (1.5%) | 163 (7.3%) | 2 (0.2%) | 1,284 (30.1%) | 438 (14.8%) | 33 (1.8%) | 8,506 (43.2%) | 65 (4.6%) |
| ‡ | |||||||||||
Clinical Profile
Patient timeline
Clinical Profile Prior to ART Initiation | Viral Load Failure 1
Plot the explanatory variable distribution for both the variables to understand the variability uniquely explained (The non-intersecting part of the blue and the red is the variation explained by the variable)
## [1] "cd4"
## [1] "hgb"
## [1] "weight"
## [1] "BMI"
## [1] "numARVs"
| Table 6: General Clinical Profile Prior to ART Initiation | Viral Load Failure 1 | ||||||
| First Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 38,268 |
 |
No No. 32,723 |
Yes No. 5,545 |
 | P-value | |
| weight | < 0.0001 | |||||
|   Mean (SD) | 58.7 (±11.6) |  | 59.0 (±11.6) | 56.8 (±11.0) |  | |
| Â Â Median (IQR) | 57.0 (51.0 - 65.0) | Â | 58.0 (51.0 - 65.0) | 55.5 (50.0 - 63.0) | Â | |
| Â Â Missing | 2,733 (7.1%) | Â | 2,297 (7.0%) | 436 (7.9%) | Â | |
| Weight Loss | < 0.0001 | |||||
| Â Â No | 35,822 (93.6%) | Â | 30,824 (86.0%) | 4,998 (14.0%) | Â | |
| Â Â Yes | 2,446 (6.4%) | Â | 1,899 (77.6%) | 547 (22.4%) | Â | |
| BMI | < 0.0001 | |||||
|   Mean (SD) | 21.8 (±4.2) |  | 21.9 (±4.2) | 21.0 (±3.9) |  | |
| Â Â Median (IQR) | 21.1 (19.0 - 23.8) | Â | 21.2 (19.1 - 23.9) | 20.4 (18.3 - 22.9) | Â | |
| Â Â Missing | 8,721 (22.8%) | Â | 7,464 (22.8%) | 1,257 (22.7%) | Â | |
| hemoglobin | < 0.0001 | |||||
|   Mean (SD) | 12.3 (±2.7) |  | 12.4 (±2.7) | 11.8 (±2.4) |  | |
| Â Â Median (IQR) | 12.5 (10.8 - 14.0) | Â | 12.5 (10.8 - 14.0) | 11.9 (10.1 - 13.4) | Â | |
| Â Â Missing | 32,862 (85.9%) | Â | 27,979 (85.5%) | 4,883 (88.1%) | Â | |
| CD4 | < 0.0001 | |||||
|   Mean (SD) | 353.5 (±266.9) |  | 366.5 (±268.6) | 267.7 (±238.4) |  | |
| Â Â Median (IQR) | 313.0 (150.0 - 492.0) | Â | 327.0 (165.0 - 503.0) | 202.0 (79.6 - 396.1) | Â | |
| Â Â Missing | 19,173 (50.1%) | Â | 16,139 (49.3%) | 3,034 (54.7%) | Â | |
| Cough | < 0.0001 | |||||
| Â Â No | 35,996 (94.1%) | Â | 30,863 (85.7%) | 5,133 (14.3%) | Â | |
| Â Â Yes | 2,272 (5.9%) | Â | 1,860 (81.9%) | 412 (18.1%) | Â | |
| Night Sweats | < 0.0001 | |||||
| Â Â No | 37,376 (97.7%) | Â | 32,027 (85.7%) | 5,349 (14.3%) | Â | |
| Â Â Yes | 892 (2.3%) | Â | 696 (78.0%) | 196 (22.0%) | Â | |
| WHO stage | < 0.0001 | |||||
| Â Â 1 | 16,271 (42.5%) | Â | 14,449 (88.8%) | 1,822 (11.2%) | Â | |
| Â Â 4 | 1,056 (2.8%) | Â | 853 (80.8%) | 203 (19.2%) | Â | |
| Â Â 2 | 6,651 (17.4%) | Â | 5,720 (86.0%) | 931 (14.0%) | Â | |
| Â Â 3 | 4,503 (11.8%) | Â | 3,648 (81.0%) | 855 (19.0%) | Â | |
| Â Â Missing | 9,787 (25.6%) | Â | 8,053 (82.3%) | 1,734 (17.7%) | Â | |
| Staging: Symptomatic HIV-associated Nephropathy/Cardiomyopathy | 1.0 | |||||
| Â Â No | 38,261 (100.0%) | Â | 32,717 (85.5%) | 5,544 (14.5%) | Â | |
| Â Â Yes | 7 (0.0%) | Â | 6 (85.7%) | 1 (14.3%) | Â | |
| TB sign: none | < 0.0001 | |||||
| Â Â No | 19,281 (50.4%) | Â | 17,014 (88.2%) | 2,267 (11.8%) | Â | |
| Â Â Yes | 18,987 (49.6%) | Â | 15,709 (82.7%) | 3,278 (17.3%) | Â | |
| TB sign: cough | < 0.0001 | |||||
| Â Â No | 37,093 (96.9%) | Â | 31,836 (85.8%) | 5,257 (14.2%) | Â | |
| Â Â Yes | 1,175 (3.1%) | Â | 887 (75.5%) | 288 (24.5%) | Â | |
| TB sign: weight loss | 0.013 | |||||
| Â Â No | 37,910 (99.1%) | Â | 32,434 (85.6%) | 5,476 (14.4%) | Â | |
| Â Â Yes | 358 (0.9%) | Â | 289 (80.7%) | 69 (19.3%) | Â | |
| ‡ | ||||||
| Table 7: Clinical Diagnosis Profile Prior to ART Initiation | Viral Load Failure 1 | ||||||
| First Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 38,268 |
 |
No No. 32,723 |
Yes No. 5,545 |
 | P-value | |
| Diagnosis: recurrent severe bacterial pneumonia | 1.00 | |||||
| Â Â No | 38,267 (100.0%) | Â | 32,722 (85.5%) | 5,545 (14.5%) | Â | |
| Â Â Yes | 1 (0.0%) | Â | 1 (100.0%) | 0 (0.0%) | Â | |
| Diagnosis: Recurrent Septicemia including non-typhoidal Salmonella | 1.0 | |||||
| Â Â No | 38,268 (100.0%) | Â | 32,723 (85.5%) | 5,545 (14.5%) | Â | |
| Â Â Yes | 0 (0.0%) | Â | 0 (%) | 0 (%) | Â | |
| Diagnosis: HIV Wasting Syndrome | 0.0008 | |||||
| Â Â No | 38,245 (99.9%) | Â | 32,710 (85.5%) | 5,535 (14.5%) | Â | |
| Â Â Yes | 23 (0.1%) | Â | 13 (56.5%) | 10 (43.5%) | Â | |
| Diagnosis: Toxoplasmosis, CNS | 0.71 | |||||
| Â Â No | 38,207 (99.8%) | Â | 32,672 (85.5%) | 5,535 (14.5%) | Â | |
| Â Â Yes | 61 (0.2%) | Â | 51 (83.6%) | 10 (16.4%) | Â | |
| Diagnosis: Extrapulmonary TB | 0.19 | |||||
| Â Â No | 38,226 (99.9%) | Â | 32,690 (85.5%) | 5,536 (14.5%) | Â | |
| Â Â Yes | 42 (0.1%) | Â | 33 (78.6%) | 9 (21.4%) | Â | |
| Diagnosis: Pulmonary TB | 0.001 | |||||
| Â Â No | 37,596 (98.2%) | Â | 32,179 (85.6%) | 5,417 (14.4%) | Â | |
| Â Â Yes | 672 (1.8%) | Â | 544 (81.0%) | 128 (19.0%) | Â | |
| Diagnosis: Lymphoma (cerebral or B-cell non-Hodgkin) | 0.37 | |||||
| Â Â No | 38,265 (100.0%) | Â | 32,721 (85.5%) | 5,544 (14.5%) | Â | |
| Â Â Yes | 3 (0.0%) | Â | 2 (66.7%) | 1 (33.3%) | Â | |
| Diagnosis: Kaposi’s Sarcoma | 0.0005 | |||||
| Â Â No | 38,168 (99.7%) | Â | 32,651 (85.5%) | 5,517 (14.5%) | Â | |
| Â Â Yes | 100 (0.3%) | Â | 72 (72.0%) | 28 (28.0%) | Â | |
| Diagnosis: Cervical Cancer | 1.0 | |||||
| Â Â No | 38,266 (100.0%) | Â | 32,721 (85.5%) | 5,545 (14.5%) | Â | |
| Â Â Yes | 2 (0.0%) | Â | 2 (100.0%) | 0 (0.0%) | Â | |
| Diagnosis: Extrapulmonary Cryptococcosis/Meningitis | 0.029 | |||||
| Â Â No | 38,128 (99.6%) | Â | 32,613 (85.5%) | 5,515 (14.5%) | Â | |
| Â Â Yes | 140 (0.4%) | Â | 110 (78.6%) | 30 (21.4%) | Â | |
| Diagnosis: Cryptosporidiosis | 0.21 | |||||
| Â Â No | 38,262 (100.0%) | Â | 32,719 (85.5%) | 5,543 (14.5%) | Â | |
| Â Â Yes | 6 (0.0%) | Â | 4 (66.7%) | 2 (33.3%) | Â | |
| Diagnosis: Cytomegalovirus Disease | 0.14 | |||||
| Â Â No | 38,267 (100.0%) | Â | 32,723 (85.5%) | 5,544 (14.5%) | Â | |
| Â Â Yes | 1 (0.0%) | Â | 0 (0.0%) | 1 (100.0%) | Â | |
| Diagnosis: HIV Encephalopathy | 0.097 | |||||
| Â Â No | 38,260 (100.0%) | Â | 32,718 (85.5%) | 5,542 (14.5%) | Â | |
| Â Â Yes | 8 (0.0%) | Â | 5 (62.5%) | 3 (37.5%) | Â | |
| Diagnosis: Candidiasis (Oesophageal, Bronchi, Trachea, or Lungs) | 0.37 | |||||
| Â Â No | 38,103 (99.6%) | Â | 32,586 (85.5%) | 5,517 (14.5%) | Â | |
| Â Â Yes | 165 (0.4%) | Â | 137 (83.0%) | 28 (17.0%) | Â | |
| Diagnosis: Chronic Herpes Simplex | 1.0 | |||||
| Â Â No | 38,252 (100.0%) | Â | 32,709 (85.5%) | 5,543 (14.5%) | Â | |
| Â Â Yes | 16 (0.0%) | Â | 14 (87.5%) | 2 (12.5%) | Â | |
| Diagnosis: Pneumocystis Pneumonia | 0.15 | |||||
| Â Â No | 38,244 (99.9%) | Â | 32,705 (85.5%) | 5,539 (14.5%) | Â | |
| Â Â Yes | 24 (0.1%) | Â | 18 (75.0%) | 6 (25.0%) | Â | |
| Diagnosis: Progressive Multifocal Leukoencephalopathy | 0.37 | |||||
| Â Â No | 38,265 (100.0%) | Â | 32,721 (85.5%) | 5,544 (14.5%) | Â | |
| Â Â Yes | 3 (0.0%) | Â | 2 (66.7%) | 1 (33.3%) | Â | |
| ‡ | ||||||
| Table 8: Prophylaxis Profile Prior to ART Initiation | Viral Load Failure 1 | ||||||
| First Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 38,268 |
 |
No No. 32,723 |
Yes No. 5,545 |
 | P-value | |
| anti-TB med: Rifabutin | 1.0 | |||||
| Â Â No | 38,265 (100.0%) | Â | 32,720 (85.5%) | 5,545 (14.5%) | Â | |
| Â Â Yes | 3 (0.0%) | Â | 3 (100.0%) | 0 (0.0%) | Â | |
| anti-TB med: Rifafour(RHZE) | < 0.0001 | |||||
| Â Â No | 37,346 (97.6%) | Â | 31,999 (85.7%) | 5,347 (14.3%) | Â | |
| Â Â Yes | 922 (2.4%) | Â | 724 (78.5%) | 198 (21.5%) | Â | |
| anti-TB med: Rifampicin(R) | 0.091 | |||||
| Â Â No | 38,036 (99.4%) | Â | 32,534 (85.5%) | 5,502 (14.5%) | Â | |
| Â Â Yes | 232 (0.6%) | Â | 189 (81.5%) | 43 (18.5%) | Â | |
| anti-TB med: Rifater(RHZ) | 0.045 | |||||
| Â Â No | 38,258 (100.0%) | Â | 32,717 (85.5%) | 5,541 (14.5%) | Â | |
| Â Â Yes | 10 (0.0%) | Â | 6 (60.0%) | 4 (40.0%) | Â | |
| anti-TB med: Rifinah(RH) | 0.0003 | |||||
| Â Â No | 38,131 (99.6%) | Â | 32,622 (85.6%) | 5,509 (14.4%) | Â | |
| Â Â Yes | 137 (0.4%) | Â | 101 (73.7%) | 36 (26.3%) | Â | |
| anti-TB med: Streptomycin | 0.12 | |||||
| Â Â No | 38,246 (99.9%) | Â | 32,707 (85.5%) | 5,539 (14.5%) | Â | |
| Â Â Yes | 22 (0.1%) | Â | 16 (72.7%) | 6 (27.3%) | Â | |
| anti-TB med: 3-FDC(RHE) | 0.42 | |||||
| Â Â No | 38,255 (100.0%) | Â | 32,713 (85.5%) | 5,542 (14.5%) | Â | |
| Â Â Yes | 13 (0.0%) | Â | 10 (76.9%) | 3 (23.1%) | Â | |
| anti-TB med: Ethambutol(E) | 0.27 | |||||
| Â Â No | 38,261 (100.0%) | Â | 32,718 (85.5%) | 5,543 (14.5%) | Â | |
| Â Â Yes | 7 (0.0%) | Â | 5 (71.4%) | 2 (28.6%) | Â | |
| anti-TB med: Ethambutol(E) | 0.11 | |||||
| Â Â No | 38,040 (99.4%) | Â | 32,537 (85.5%) | 5,503 (14.5%) | Â | |
| Â Â Yes | 228 (0.6%) | Â | 186 (81.6%) | 42 (18.4%) | Â | |
| anti-TB med: INH | 1.0 | |||||
| Â Â No | 38,227 (99.9%) | Â | 32,688 (85.5%) | 5,539 (14.5%) | Â | |
| Â Â Yes | 41 (0.1%) | Â | 35 (85.4%) | 6 (14.6%) | Â | |
| anti-TB med: Other | 1.00 | |||||
| Â Â No | 38,267 (100.0%) | Â | 32,722 (85.5%) | 5,545 (14.5%) | Â | |
| Â Â Yes | 1 (0.0%) | Â | 1 (100.0%) | 0 (0.0%) | Â | |
| anti-TB med: Pyrazinamide(Z) | 0.11 | |||||
| Â Â No | 38,038 (99.4%) | Â | 32,535 (85.5%) | 5,503 (14.5%) | Â | |
| Â Â Yes | 230 (0.6%) | Â | 188 (81.7%) | 42 (18.3%) | Â | |
| TB Prophylaxis: INH | 0.67 | |||||
| Â Â No | 36,568 (95.6%) | Â | 31,263 (85.5%) | 5,305 (14.5%) | Â | |
| Â Â Yes | 1,700 (4.4%) | Â | 1,460 (85.9%) | 240 (14.1%) | Â | |
| OI Prophylaxis | < 0.0001 | |||||
| Â Â No | 9,966 (26.0%) | Â | 8,937 (89.7%) | 1,029 (10.3%) | Â | |
| Â Â Yes | 28,302 (74.0%) | Â | 23,786 (84.0%) | 4,516 (16.0%) | Â | |
| TB treatment | 0.012 | |||||
| Â Â No | 37,191 (97.2%) | Â | 31,831 (85.6%) | 5,360 (14.4%) | Â | |
| Â Â Yes | 1,077 (2.8%) | Â | 892 (82.8%) | 185 (17.2%) | Â | |
| ‡ | ||||||
| Table 9: ARVs Profile Prior to ART Initiation | Viral Load Failure 1 | ||||||
| First Virologic Failure | ||||||
|---|---|---|---|---|---|---|
|
Total No. 38,268 |
 |
No No. 32,723 |
Yes No. 5,545 |
 | P-value | |
| # current ARVs | < 0.0001 | |||||
|   Mean (SD) | 3.0 (±0.2) |  | 3.0 (±0.2) | 3.0 (±0.3) |  | |
| Â Â Median (IQR) | 3.0 (3.0 - 3.0) | Â | 3.0 (3.0 - 3.0) | 3.0 (3.0 - 3.0) | Â | |
| Â Â Missing | 3,057 (8.0%) | Â | 2,460 (7.5%) | 597 (10.8%) | Â | |
| ARVs taken last 7 days | 0.21 | |||||
| Â Â about half | 0 (0.0%) | Â | 0 (%) | 0 (%) | Â | |
| Â Â all | 1,515 (4.0%) | Â | 1,255 (82.8%) | 260 (17.2%) | Â | |
| Â Â most | 3 (0.0%) | Â | 2 (66.7%) | 1 (33.3%) | Â | |
| Â Â none | 94 (0.2%) | Â | 78 (83.0%) | 16 (17.0%) | Â | |
| Â Â very few | 1 (0.0%) | Â | 0 (0.0%) | 1 (100.0%) | Â | |
| Â Â Missing | 36,655 (95.8%) | Â | 31,388 (85.6%) | 5,267 (14.4%) | Â | |
| ART adherence: percent taken since last visit | 0.14 | |||||
|   Mean (SD) | 98.9 (±8.4) |  | 99.5 (±4.1) | 95.5 (±18.7) |  | |
| Â Â Median (IQR) | 100.0 (100.0 - 100.0) | Â | 100.0 (100.0 - 100.0) | 100.0 (100.0 - 100.0) | Â | |
| Â Â Missing | 37,871 (99.0%) | Â | 32,389 (99.0%) | 5,482 (98.9%) | Â | |
| ARV Adherence- combines formats from both the old and the new CTC 2 | 1.0 | |||||
| Â Â Good: fewer than 2 missed days | 322 (0.8%) | Â | 289 (89.8%) | 33 (10.2%) | Â | |
| Â Â Poor: 2 or more missed days | 4 (0.0%) | Â | 4 (100.0%) | 0 (0.0%) | Â | |
| Â Â Missing | 37,942 (99.1%) | Â | 32,430 (85.5%) | 5,512 (14.5%) | Â | |
| ARV adherence since last visit | < 0.0001 | |||||
| Â Â FAIR (85-94%) | 63 (0.2%) | Â | 44 (69.8%) | 19 (30.2%) | Â | |
| Â Â GOOD (>, 95%) | 5,953 (15.6%) | Â | 5,091 (85.5%) | 862 (14.5%) | Â | |
| Â Â POOR (< 85%) | 86 (0.2%) | Â | 53 (61.6%) | 33 (38.4%) | Â | |
| Â Â Missing | 32,166 (84.1%) | Â | 27,535 (85.6%) | 4,631 (14.4%) | Â | |
| ‡ | ||||||
Bivariate Plots
To understand the relationship of each continuous variable with the y variable:
- Plot box plot for each of the variables to do a visual comparison between the groups
- Plot the explanatory variable distribution for both the variables to understand the variability uniquely explained (The non-intersecting part of the blue and the red is the variation explained by the variable)
Socio-demographics
## [1] "Height"
## [1] "Weight"
## [1] "bmi"
## [1] "ageatarvstart"
## [1] "ageatarvstartTHERAP"
## [1] "num_days_to_vl1"
## [1] "num_days_to_vl2"
## [1] "num_days_to_vl3"
## [1] "vl_count1"
## [1] "vl_count2"
## [1] "vl_count3"
## [1] "num_encs_to_vl1"
## [1] "num_encs_to_vl2"
## [1] "num_encs_to_vl3"
Univariate Analysis
For each continuous column, visually check the following:
- Variation in the column
- Its distribution
- Any outliers
- q-q plot with normal distribution
## [1] "Height"
## [1] "Weight"
## [1] "bmi"
## [1] "ageatarvstart"
## [1] "ageatarvstartTHERAP"
## [1] "num_days_to_vl1"
## [1] "num_days_to_vl2"
## [1] "num_days_to_vl3"
## [1] "vl_count1"
## [1] "vl_count2"
## [1] "vl_count3"
## [1] "num_encs_to_vl1"
## [1] "num_encs_to_vl2"
## [1] "num_encs_to_vl3"
For categorical variables,visually check the frequencies and percentages.
## [1] "vl_failure2" "vl_failure1" "male" "Death"
## [1] "civilstatusENROL" "transfer" "entrypoint" "nhif"